Project Description With growing awareness of how pathogen adaptation impacts the battle against infectious disease, mathematical models of adaptation have become central to this fight. However, most of the theoretical work focuses on general patterns of adaptation, while the empirical work provides rich details specific to the pathogen under study. For those who wish to predict adaptation of a specific pathogen, the extensive biological information cannot be easily incorporated into existing models. The long term goal of this research is to develop a flexible framework for predicting evolution that is rich enough to accommodate empirical data from organisms that evolve in real time. The 'GPF' model proposed here builds on the knowledge that genotypes (G) affect phenotypes (P) and phenotypes affect fitness (F). This framework traces back to Fisher's geometric model, which serves as a baseline for comparison. There are three Aims. Aim 1: Test three key assumptions of the geometric model on viral phenotypes and fitness. In the G to P part of the GPF model, the assumptions that mutations show universal pleiotropy at the phenotype level and that phenotypic effects of mutations are additive at the phenotype level are tested. In mapping P to F, the model departs from the standard assumption of a multidimensional Gaussian function by allowing the relationship to emerge from a basic life-cycle model with observable phenotypes as predictors of fitness. These assumptions will be tested using a well-developed viral model system for which a large library of previously observed adaptive mutations is available. A subset of mutations will be engineered into single, double, and triple mutation combinations and assaying each at six phenotypic traits plus fitness. Aim 2: Synthesize the results of Aim 1 into a unified model, make predictions about adaptations and test them. Biologically reasonable modifications will be evaluated through model selection. Mathematical simulations under the refined GPF model will be used to make quantitative predictions about important general properties of adaptive walks, and these properties will be tested by carrying out adaptation in the laboratory. The model will be evaluated based on how close predictions match observed data. Aim 3: Use the unified model to design genomes and test predicted fitness. In this Aim, the GPF model will be refocused from general patterns to specific predictions about the phenotypes and fitnesses. Multistep genotypes will be engineered from the single mutations tested in Aim 1, their phenotypes and fitnesses assayed, and the results compared to predictions. Next, the growth environment will be altered in a specific way and the GPF model will have the more challenging task of predicting what multistep genotypes will have high fitness in the novel environment. These genotypes will be engineered, and their fitness assayed and compared to the GPF predictions and to laboratory adaptations in the novel environment. Finally, the predictive successes and failures will be critically evaluated to shed light on how future research can advance the larger goal of producing a predictive model of microbial evolution useful to the study of human pathogens. PUBLIC HEALTH RELEVANCE: Project Narrative Many pathogens evolve in real time, enabling them to become resistant to our drugs and vaccines or to jump to new hosts. A predictive model of pathogen evolution would be immensely useful in combating infectious diseases. The proposed research will mark a significant step toward this end by developing and testing a more biologically realistic model of adaptation through strong integration of theory and experimental work.